| Acronym | Name |
|---|---|
| TE | Evergreen trees |
| TDdry | Drought-deciduous trees |
| TDcold | Cold-deciduous trees |
| TN | Needleleaf trees |
| ShE | Evergreen shrubs |
| ShDdry | Drought-deciduous shrubs |
| ShDcold | Cold-deciduous shrubs |
| H | Herbs |
| Geo | Geophytes |
| Thero | Therophytes |
| GC3 | C3 grasses |
| GC4 | C4 grasses |
| Suc | Succulents |
| Clim | Climbers |
Velocity metrics
The setup.
Here, I will be exploring how the velocity based novelty metrics in (Ordonez, Williams, and Svenning 2016) can be used in (Conradi et al. 2020) work on Phyto-climates that builds on his work on operationalizing the definition of the biome for global change research.
The novelty metrics in (Ordonez, Williams, and Svenning 2016) focus on measuring three different mechanisms by which ecological novelty might emerge. These mechanisms are based on the idea that as environmental changes happen the composition of taxa on a site will change change.
Our goals are:
Developing new metrics of ecosystem change based on velocity vectors of phytoclimatic change.
Provide a novel and nuanced perspective to identify the ecosystems most at risk from climate change.
Input information.
Here using will use the phytoclimates maps of 14 growth form (GF):
Data Inputs - Phytoclimates.
All analyses use Phytoclimates as inputs. These variables can be defined as summaries of the climatic suitability for all plant species with a given growth form (e.g. evergreen tree, grass). Its estimation is based on a physiologically informed suitability model (i.e., an Eco-physiological species distribution Model - TTR). The models are based on defining the functional form of physiological constraints to plant growth (as modelled in Dynamic Global Vegetation Models [DGVMs]), parameterizing these using occurrence data, and then using the statistical methods of Species Distribution Modelling (SDMs) to define suitability surfaces for each evaluated species..
The values in the figures above (i.e., phytoclimate suitability) can be defined as the proportion of the species within a growth form for which the environmental conditions at 50x50km a grid-cell are considered suitable at a given period (here 1950; Figure 1).
What is the velocity of phytoclimatic change?
The Velocity of environmental change idea is built on the approach developed by (Loarie et al. 2009), where velocity for an environmental variable (e.g., temperature) is estimated as:
\[V_{l} = \frac{\text{d}c/\text{d}t}{\text{d}c/\text{d}x}\]
Where \(\frac{\text{d}c}{\text{d}t}\) is the ratio between the projected change per unit time, and \(\frac{\text{d}c}{\text{d}x}\) is the local spatial gradient in the variable of interest.
In this context, velocity vectors for phytoclimates measure the rapidity (speed) and direction (bearing) of a change in the suitability of environmental conditions for the species in a given growth form. You can think of this as a measurement of how fast , and in which direction would the “suitability” surface of an average species in a given growth form has/will move in space (like in (Serra-Diaz et al. 2014)).
This metric answer the question, “How fast to move to keep the same ‘suitability’”, assuming that the movement is from an area of low to a place of higher suitability between two times points. A 10km/year value means that to compensate for the change in 1yr, you would need to move to a location 10km away. I apply this approach to each phytoclimates suitability map rather than a single climate variable.
The approach used to estimate the velocity of phytoclimatic change (that is, the magnitude and direction of the change vector) flows the implementation in (Ordonez et al. 2014) and (Ordonez, Williams, and Svenning 2016).
Maping the speed of phytoclimatic change.
The magnitude of Phytoclimate velocity vectors (i.e., speed; Figure 2) was the fastest in most of the northern hemisphere, particularly those areas covered by ice sheets during the LGM (the Cordillera, Laurentide, Innuitian, Greenland, Barents-Kara, Fennoscandia and British-Irish Ice Sheets; see (Ehlers, Gibbard, and Hughes 2011)). most likely due to Large temporal changes in suitability.
The Sahara and central and western Australia regions also show large velocities for all phytoclimates (Figure 2), particularly those from drought-deciduous shrubs(ShDdry), drought-deciduous trees (TDdry), Therophytes (Thero), and needle leaf trees (TN) growth forms. Such fast speeds are likely due to low suitability for these phytoclimates and due to homogeneous spatial patterns.
Fastest changing phytoclimate.
When comparing the velocity between phytoclimates in each cell (i.e., suitability for the avg species in growth forms), Climbers (Clim) were the group that showed, in most cases, the fastest speeds for a grid (Figure 3). If Climbers are removed from the analysis, C3 and C4 grasses become the growth forms with the fastest velocities (Figure 4).
Maping the displacement of phytoclimate velocity vectors.
A compound metric of change that shows the average rapidity in the response across multiple variables is the displacement of velocity vectors (cf. (Ordonez, Williams, and Svenning 2016)). It is a metric of how fast you would need to move the keep the same environmental setup.
NOTE: Could this be considered as to how fast a given phytoclimate assemblage (defined by its suitability) must move in response to environmental changes?
Using the phytoclimate velocity vectors, these displacement is estimated (Figure 5) using the suitability composition for all phytoclimates in a location.
This map shows again that the displacement of phytoclimate velocity vectors (Figure 5) was the fastest in most of the northern hemisphere, particularly those areas covered by ice sheets during the LGM (the Cordillera, Laurentide, Innuitian, Greenland, Barents-Kara, Fennoscandia and British-Irish Ice Sheets; see (Ehlers, Gibbard, and Hughes 2011)).
Sensitivity of displacement estimates.
The displacement of phytoclimates showed the highest sensitivity in most evaluated cells to the speed of drought-deciduous shrubs (ShDdry) (Figure 6). This sensitivity was estimated as the absolute difference between displacement estimate with all phytoclimates and the estimation after removing a given phytoclimate.
A second group of phytoclimates with a strong influence on displacement estimates (Figure 6) were Cold-deciduous trees (TDcold), Needle-leaf trees (TN), Evergreen trees (TE), and Drought-deciduous trees (TDdry). These were the most influential phytoclimates in a similar number of cells.
Displacement: Mean vs Variability.
It is also relevant to consider the balance between the average and variability in the rapidity of velocity vectors (Figure 7). These show that for most areas, there is an overall consistency in the velocity of phytoclimates (for most cells, the variability is low (median variability ~7.73km/100yrs), a value that is comparable to the median velocity of phytoclimates (~3.1km/100yrs).
Maping the bearing of phytoclimatic vectors.
A second dimension of velocity vectors is their direction, which is the angle measured counter-clockwise from north. Here I adjusted these so they represent poleward/equatorward directions (0-degrees is pole ward, 180 degree is equatorward). in this way, bearing measurement show show the movement from the tropics to temperate to polar areas.
For most phytoclimates, vectors show a Poleward direction (Figure 8), aligned with the idea of a colonization of post-glaciated areas. However, for the large areas covered by the the Cordillera, Laurentide, Innuitian, Greenland, Barents-Kara, Fennoscandia and British-Irish Ice Sheets; see (Ehlers, Gibbard, and Hughes 2011)) is not possible to estimate the bearing due to homogeneous/flat spatial gradients.
Maping the divergence of phytoclimate velocity vectors.
Differences in the bearing of phytoclimate velocity vectors were small overall as shown by the overall low median angle between phytoclimate velocity vectors (i.e., divergence; (Figure 9)). In those areas where bearings did nit converge, there is a general trend towards orthogonal bearings (e.g., Cerrado region, Northern and central Africa, Borneo, Eastern Australia; see (Figure 9)), indicating a possible dissociation in the suitability fo different phytoclimates.
Bearing Mean vs variability
To further test the idea that bearings of phytoclimates show similar directions, I plot the mean vs the variability in the angles between pairwise phytoclimatic velocity vectors. These show that for some cells there is a large variability in the match in directions between vectors (Figure 10 A), for the vast majority of locations divergence is low (>60 degrees), and with a low variability (>225. degrees) (Figure 10 B).